A major impulse for the study of subjective well-being has been the question how policy affects the distribution of subjective well-being within a country. A natural approach to learning about policy effects is to compare policies across countries and relate these to observed differences in well-being. Most of the work on international comparisons of well-being has been based on global satisfaction measures, like single questions inviting respondents to rate their happiness on a numerical or verbal scale. This work has produced widely conflicting findings. More recently, experienced well-being has been proposed as an alternative measure. Both approaches to the measurement of well-being are in principle vulnerable to Differential Item Functioning, particularly when used to compare different countries. We propose a thorough analysis of the properties of the different well-being measures and in particular how anchoring vignettes can be used to improve the comparability of well-being measures across population groups and across countries. To that end we will conduct several experiments in two population representative Internet panels: the American Life Panel (ALP) in the U.S. and the Longitudinal Internet Studies for the Social sciences (LISS) panel in The Netherlands.